Re: [coreboot] Change in coreboot[master]: dmp/vortex86ex: Initialize I2C controller base address/IRQ

[Paul Menzel]

Dear Andrew,

let’s move the discussion of [1] to the list.

Am Mittwoch, den 23.10.2013, 13:43 +0200 schrieb gerrit code review:
> Change subject: dmp/vortex86ex: Initialize I2C controller base address/IRQ
> ......................................................................
>
> Patch Set 5:
>
> Do you mean allocating I2C I/O space range dynamically, instead of
> using fixed address? It makes sense, but :
>
> (A) I don't know if coreboot has an API for I/O space
> allocation/reservation. If yes, could you tell me the function names
> so I can check it.
>
> (B) Some legacy applications for our CPU maybe use fixed I/O addresses
> for these devices, because these I/O addresses are fixed in old BIOS.

hopefully, some of the other developers can answer this.

Thanks,

Paul

[1] http://review.coreboot.org/3976

Change in coreboot[master]: dmp/vortex86ex: Initialize I2C controller base address/IRQ

[Rudolf Marek]

Hi,

If Andrew wants to use the fixed address it is better to marked the fixed IO 
address in the I/O allocator as used. Like in VIA vt8237 for ACPI IO base:

static void vt8237r_read_resources(device_t dev)
{
         struct resource *res;

pci_dev_read_resources(dev);

/* Fixed ACPI Base IO Base*/
         res = new_resource(dev, 0x88);
         res->base = VT8237R_ACPI_IO_BASE;
         res->size = 128;
         res->limit = 0xffffUL;
         res->flags = IORESOURCE_IO | IORESOURCE_FIXED | IORESOURCE_RESERVE |
                      IORESOURCE_STORED | IORESOURCE_ASSIGNED;
}

But I see this dmp southbrige bypassed the I/O allocator by reserving it whole 
for southbridge.

If no PCI or I/O devices will ever connect this is probably OK, although I would 
prefer to use above construct to add "fixed" I/O resources.

I did not have time to study the code/chipset in the detail, if someone can 
check that it would be great.

Thanks
Rudolf

Change in coreboot[master]: dmp/vortex86ex: Initialize I2C controller base address/IRQ

[Andrew Wu]

Hi, Paul, Rudolf,

Thank you for the information. I will investigate this part. :D

Andrew

2013/10/28 Rudolf Marek :
> Hi,
>
> If Andrew wants to use the fixed address it is better to marked the fixed IO
> address in the I/O allocator as used. Like in VIA vt8237 for ACPI IO base:
>
> static void vt8237r_read_resources(device_t dev)
> {
>         struct resource *res;
>     pci_dev_read_resources(dev);
>
>     /* Fixed ACPI Base IO Base*/
>     res = new_resource(dev, 0x88);
>     res->base = VT8237R_ACPI_IO_BASE;
>     res->size = 128;
>     res->limit = 0xffffUL;
>     res->flags = IORESOURCE_IO | IORESOURCE_FIXED | IORESOURCE_RESERVE |
>                  IORESOURCE_STORED | IORESOURCE_ASSIGNED;
>
> }
>
> But I see this dmp southbrige bypassed the I/O allocator by reserving it
> whole for southbridge.
>
> If no PCI or I/O devices will ever connect this is probably OK, although I
> would prefer to use above construct to add "fixed" I/O resources.
>
> I did not have time to study the code/chipset in the detail, if someone can
> check that it would be great.
>
> Thanks
> Rudolf